hm-nodejs-queue

Install bullmq and ioredis, create a Queue instance with a shared Redis connection, and define a Worker with a processor function. Dispatch jobs from services with queue.add() and let workers process them asynchronously with configurable concurrency.

Use the repeat option with a cron expression when adding a job, such as repeat: { cron: "0 9 * * MON" }. On startup, remove existing repeatable jobs with removeRepeatableByKey before re-adding to prevent duplicates after restarts.

BullMQ uses blocking Redis commands that can exceed the default retry limit, causing ioredis to throw errors. Setting maxRetriesPerRequest to null and enableReadyCheck to false on the shared connection is required for BullMQ to function correctly.

Same-process workers are acceptable for small apps, but a separate worker process is preferred in production. It scales independently of the HTTP layer and prevents a worker crash from bringing down the API.

Set attempts and backoff in defaultJobOptions at the queue level, for example attempts: 3 with exponential backoff starting at 1000ms. When a worker throws, BullMQ automatically reschedules the job until attempts are exhausted, then marks it failed.

Use Bull Board with the @bull-board/fastify adapter to expose a web UI at a route like /admin/queues for inspecting and retrying jobs. In production, protect the route with authentication and admin authorization middleware.